Disease Models & Mechanisms 4, 659-672 (2011) doi:10.1242/dmm.

007419

RESEARCH ARTICLE

A Drosophila model for the Zellweger spectrum of peroxisome biogenesis disorders

Fred D. Mast1,*, Jing Li1,*, Maninder K. Virk1, Sarah C. Hughes1,2, Andrew J. Simmonds1 and Richard A. Rachubinski1,
SUMMARY Human peroxisome biogenesis disorders are lethal genetic diseases in which abnormal peroxisome assembly compromises overall peroxisome and cellular function. Peroxisomes are ubiquitous membrane-bound organelles involved in several important biochemical processes, notably lipid metabolism and the use of reactive oxygen species for detoxification. Using cultured cells, we systematically characterized the peroxisome assembly phenotypes associated with dsRNA-mediated knockdown of 14 predicted Drosophila homologs of PEX genes (encoding peroxins; required for peroxisome assembly and linked to peroxisome biogenesis disorders), and confirmed that at least 13 of them are required for normal peroxisome assembly. We also demonstrate the relevance of Drosophila as a genetic model for the early developmental defects associated with the human peroxisome biogenesis disorders. Mutation of the PEX1 gene is the most common cause of peroxisome biogenesis disorders and is one of the causes of the most severe form of the disease, Zellweger syndrome. Inherited mutations in Drosophila Pex1 correlate with reproducible defects during early development. Notably, Pex1 mutant larvae exhibit abnormalities that are analogous to those exhibited by Zellweger syndrome patients, including developmental delay, poor feeding, severe structural abnormalities in the peripheral and central nervous systems, and early death. Finally, microarray analysis defined several clusters of genes whose expression varied significantly between wild-type and mutant larvae, implicating peroxisomal function in neuronal development, innate immunity, lipid and protein metabolism, gamete formation, and meiosis.

Disease Models & Mechanisms DMM

INTRODUCTION Peroxisome biogenesis disorders (PBDs) are a broad spectrum of human diseases originating from mutations in genes encoding proteins responsible for the formation and maintenance of peroxisomes. Peroxisome number, shape, size and protein composition vary dramatically depending on the cell type and environment. In mammals, peroxisomes are particularly abundant in the liver, kidneys and central nervous system (CNS), where they are found primarily in the oligodendroglia. Peroxisomes contain more than 50 different enzymes that are involved in a variety of important metabolic pathways, including the oxidative breakdown of fatty acids, peroxide detoxification, the oxidation of L- and Damino acids and polyamines, and the synthesis of plasmalogens, bile acids, ether phospholipids and polyunsaturated fatty acids [for classical and current views of peroxisomes and their functions, see the following references (Lazarow and Fujiki, 1985; van den Bosch et al., 1992; Wanders and Waterham, 2006; Schrader and Fahimi, 2008; Mast et al., 2010)]. The contribution of peroxisome function to normal human development and physiology is underscored by the severity and lethality of PBDs. PBDs are a heterogeneous group of fatal

Department of Cell Biology, and 2Department of Medical Genetics, University of Alberta, Edmonton, AB T6G 2H7, Canada *These authors contributed equally to this work Author for correspondence (rick.rachubinski@ualberta.ca) Received 15 December 2010; Accepted 4 May 2011
2011. Published by The Company of Biologists Ltd This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial Share Alike License (http://creativecommons.org/licenses/by-nc-sa/3.0), which permits unrestricted non-commercial use, distribution and reproduction in any medium provided that the original work is properly cited and all further distributions of the work or adaptation are subject to the same Creative Commons License terms.

autosomal recessive diseases that includes Zellweger syndrome (ZS), rhizomelic chondrodysplasia punctata (RCDP), neonatal adrenoleukodystrophy (NALD) and infantile Refsum disease (IRD). These four disorders are classified on the basis of the severity of clinical phenotypes. In addition to these broad biogenic disorders, there are additional single-enzyme peroxisome diseases that share many of the same traits as these more serious disorders. In general, patients with PBD suffer from profound neurological abnormalities, muscular hypotonia, cataracts, cardiac defects, dysmorphic features, and growth and mental retardation (for reviews, see Shimozawa et al., 2005; Steinberg et al., 2006). ZS is the most severe and typically presents with severe hypotonia, such that children with ZS are sometimes initially misdiagnosed with various forms of developmental, mental and physical retardation, including Down syndrome, Prader-Willi syndrome or spinal muscular dystrophy. Patients with ZS typically die in the first year of life (Wilson et al., 1988; Yik et al., 2009). The overall incidences of the PBDs vary widely among different population groups but are invariably infrequent enough to be classified as rare genetic diseases. ZS has an estimated incidence of 1 in 50,000 in North America but approximately 1 in 500,000 in Japan (Steinberg et al., 2004). The molecular mechanisms underlying peroxisome biogenesis have been strongly conserved during evolution (Platta and Erdmann, 2007; Schrader and Fahimi, 2008). To date, 34 PEX genes encoding peroxins, or proteins required for peroxisome assembly, have been identified (Platta and Erdmann, 2007; Mast et al., 2010; Tower et al., 2011). Of these, 13 PEX genes functionally complement the defects in peroxisome assembly of all the known complementation groups of the PBDs (Shimozawa et al., 2005; Steinberg et al., 2006). These 13 PEX genes function in a variety of different steps in the peroxisome biogenic pathway, including peroxisomal membrane assembly and peroxisomal protein
659

Disease Models & Mechanisms

RESEARCH ARTICLE

Drosophila peroxisome assembly

Disease Models & Mechanisms DMM

targeting. An unexpected complication in the assignment of PBD phenotypes to any particular complementation group is the fact that different complementation groups share similar clinical phenotypes (Moser et al., 1995; Steinberg et al., 2006). This is most probably due to the fact that a defect in any aspect of the peroxisome biogenic program results in the same functional loss of the organelle. As a result, clinical diagnosis of the cause of any one PBD has required an in-depth and tedious biochemical characterization followed by sequencing efforts to pinpoint the source of the mutation (Steinberg et al., 2004; Shimozawa, 2011). Advances in sequencing technology have improved the speed of diagnosis and there is currently an undertaking to fully annotate and catalog the range of mutations present along the PBD continuum (http://dbpex.org/home.php). In addition, model systems have greatly aided the identification and characterization of peroxisomal genes (Titorenko and Rachubinski, 2001). Although single-cell systems such as yeast and mammalian cell culture have been invaluable for identifying the proteins and basic processes involved in peroxisome biogenesis, they cannot act de facto as models for the study of the role of peroxisomes in the development of multicellular organisms or in the molecular mechanisms leading to PBDs. Drosophila has been shown to be both a valid and valuable model system for the study of normal and abnormal development of more complex multicellular organisms, including humans. Over 75% of known human disease genes have a recognizable match in the genome of Drosophila (Reiter et al., 2001), and Drosophila now serves as a model for more than 20 diseases, including Huntingtons, Parkinsons, Alzheimers, fragile X mental retardation and spinocerebellar ataxia 1 (Bilen and Bonini, 2005; Cauchi and van den Heuvel, 2006; Raymond and Tarpey, 2006; Snchez-Martnez et al., 2006). Surprisingly, there have been few studies of the functional roles of peroxisomes in Drosophila and, until this study, none focused

on the effects of peroxisome function on early development. The longest-known role of peroxisomes in flies is in the formation of eye pigments. The rosy (ry) gene encodes the enzyme xanthine dehydrogenase (Bridges and Brehme, 1944; Glassman and Mitchell, 1958), which is targeted to peroxisomes in Malpighian tubule cells and fat cells (Beard and Holtzman, 1987). Recent studies have also suggested a role for peroxins in the process of spermatogenesis in adult males (Chen et al., 2010). However, no study has focused specifically on the functional roles of peroxisome biogenesis (Pex) genes of Drosophila melanogaster or on the defects in early fly development arising from mutation of a Pex gene, and the similarity between those defects and the developmental defects observed in PBD patients mutant for the corresponding human PEX gene. To first establish a Drosophila model for the embryonic development defects associated with the PBDs, we performed a systematic double stranded (ds)RNA knockdown screen in embryonically derived Schneider 2 (S2) cells to confirm homologous roles of the 14 Drosophila genes homologous to known PEX genes in other organisms (Fig. 1) (Chen et al., 2010). Importantly, 12 of the PEX genes have known mutations that lead to PBD formation. Once we had demonstrated that the majority of Drosophila Pex proteins are required for peroxisome assembly, we next assayed the functional requirement for peroxisomes in Drosophila embryonic development by examining the developmental phenotype of flies with homozygous mutations in Pex1, because mutations in human PEX1 are the most common cause of PBD (Reuber et al., 1997; Steinberg et al., 2006). Pex1 mutant larvae exhibit abnormalities that are analogous to those observed in ZS patients. These abnormalities correlate to reproducible changes in gene expression between larvae lacking functional Pex1 and wild-type larvae. Our results support the use of Drosophila for studying the role of PEX genes in early development and as a genetic model for PBDs and, in particular, ZS.

Fig. 1. Peroxins and their putative homologs in Drosophila. The putative Drosophila homologs of known peroxins that we identified in silico are presented. The main function in peroxisome biogenesis of each known peroxin is given. The PBD in which a PEX gene has been implicated is indicated in brackets. Pairwise alignment is made between the human or yeast peroxin (upper) and the putative Drosophila homolog (lower) using the SIM alignment algorithm (http://ca.expasy.org/tools/sim-prot.html) and visualized using Lalnview (http://pbil.univ-lyon1.fr/ software/lalnview.html). The extent of amino acid similarity between regions of two aligned proteins is given by the Heat map at bottom. Hs, Homo sapiens; Yl, Yarrowia lipolytica; Dmel, Drosophila melanogaster.

660

dmm.biologists.org

Drosophila peroxisome assembly

RESEARCH ARTICLE

RESULTS RNAi analysis in S2 cells confirms that the majority of the known genes required for peroxisome assembly are conserved in Drosophila Fifteen putative Pex genes are predicted in the Drosophila genome. Thirteen genes are homologous to known human PEX genes, whereas two genes are homologous to the PEX20 and PEX23 genes of the yeast Yarrowia lipolytica (Fig. 1). These Drosophila Pex genes correlate with genes predicted by others as Drosophila homologs for PEX genes found in other species (Adams et al., 2000; Chen et al., 2010). Twelve of these genes are homologous to known human PEX genes for which mutations causing PBDs have been characterized. Mutations in PEX genes typically result in the mislocalization of peroxisomal matrix proteins to the cytosol, and this mislocalization of matrix proteins is used along with other biochemical laboratory diagnostics in characterizing a PBD (Shimozawa et al., 1999; Steinberg et al., 2006; Shimozawa, 2011). We performed systematic dsRNA interference (RNAi) knockdown

of each Pex gene in S2 cells constitutively expressing a chimeric reporter protein, GFP-SKL. The SKL tripeptide is the evolutionarily conserved C-terminal peroxisome targeting signal 1 (PTS1), consisting of Ser-Lys-Leu (SKL) that targets specifically to peroxisomes to produce a characteristic punctate pattern in fluorescence microscopy (Petriv et al., 2002; Kural et al., 2005). GFPSKL-expressing S2 cells were treated with dsRNAs targeting 14 individual putative Drosophila Pex genes for knockdown: Pex1 (CG6760), Pex2 (CG7081), Pex3 (CG6859), Pex5 (CG14815), Pex6 (CG11919), Pex7 (CG6486), Pex11 (CG8315), Pex12 (CG3639), Pex13 (CG4663), Pex14 (CG4289), Pex16 (CG3947), Pex19 (CG5325), Pex20 (CG3696) and Pex23 (CG32226) (Fig. 1). As a control for specificity of the RNAi knockdown, we performed a mock RNAi treatment (no dsRNA), as well as an RNAi knockdown of Dredd, which has no known role in peroxisome biogenesis or function. Untreated GFP-SKL S2 cells exhibited a punctate pattern of fluorescence, which is characteristic of peroxisomes (Fig. 2) (Kural et al., 2005). Similarly, in mock-treated and Dredd RNAi-treated

Disease Models & Mechanisms DMM

Fig. 2. Peroxisomes are absent or exhibit altered morphology in S2 cells treated with dsRNA to putative Pex genes. (A)S2 cells constitutively expressing the fluorescent peroxisomal reporter protein GFP-SKL (Kural et al., 2005) were treated with dsRNA to the indicated putative Pex genes, mocktreated (MT) or treated with dsRNA targeting Dredd, which has no known role in peroxisome biogenesis or peroxisome function. GFP-SKL in S2 cells targets to punctae that are characteristic of peroxisomes. Mock-treated cells and cells treated with dsRNA targeting Dredd exhibited punctae in the same pattern as control cells (UT). Cells treated with dsRNAs to different Pex genes exhibited mislocalization of the GFP-SKL peroxisomal reporter to the cytosol and/or altered peroxisomal size and number. Cells treated with dsRNA to Pex7 or Pex20 exhibited punctate peroxisomes that were essentially like those of wild-type cells, because PEX7 and PEX20 affect the targeting only of peroxisomal proteins containing a PTS2 and not of those containing PTS1, such as GFP-SKL. (B) Quantitative description of peroxisome morphologies in S2 cells treated with dsRNA to putative Pex genes. Images were scored for numbers of punctate bodies, the average volume of these punctate bodies and the intensity of fluorescent signal from these punctate bodies.

Disease Models & Mechanisms

661

RESEARCH ARTICLE

Drosophila peroxisome assembly

cells, the GFP-SKL signal also appeared in punctate structures, consistent with peroxisome localization. The peroxisomes in control and untreated cells were spherical, of essentially uniform size and randomly distributed throughout the cell. In S2 cells, RNAi knockdown of many of the putative Drosophila Pex genes that we identified (supplementary material Fig. S1 and Table S1) resulted in GFP-SKL fluorescence patterns that were significantly different from the punctate fluorescence pattern observed in untreated or control-treated cells (Fig. 2A). We grouped these patterns into five classes (Fig. 2B). Class 1 cells exhibit a pattern of GFP-SKL fluorescence that is essentially indistinguishable from that of untreated cells. Mock-treated and Dredd RNAi-treated cells were categorized as Class 1, as were cells treated with dsRNA targeting Pex7. This was not unexpected, because Pex7p has been shown to function in the targeting to peroxisomes of proteins containing an N-terminally localized PTS2 but not of proteins containing a C-terminally localized PTS1 (Lazarow, 2006) (i.e. the chimeric reporter GFP-SKL being expressed in these S2 cells). In Class 2 cells, most of the GFP signal was excluded from punctate bodies. Putative Pex genes in Class 2 include Pex1, Pex5, Pex13 and Pex16. Class 3 cells show an appreciable amount of cytosolic GFP signal but contain fluorescent punctate bodies in numbers approaching those of untreated cells and were observed in RNAi knockdown of Pex2, Pex3, Pex6, Pex12 and Pex14. Class 4 cells have a reduced number of fluorescent punctate bodies of increased size as compared with untreated cells but show no appreciable accumulation of GFP signal in the cytosol. RNAi knockdown of the Pex11 and Pex19 genes produced a Class 4 pattern. In Class 5 cells, the average volume of punctate bodies was less and the average number of punctate bodies per cell was greater than that of untreated cells, and all the GFP signal was essentially localized to punctate bodies. Cells treated with dsRNA to Pex20 or Pex23 were categorized as Class 5. The specificity of transcript knockdown by RNAi was verified for the case of Pex1 by semi-quantitative reverse-transcriptase PCR (RT-PCR; Fig. 3A) and by immunoblotting of whole S2 cell lysates with antibody to Pex1 protein (Fig. 3B). The levels of Pex1 mRNA were reduced in GFP-SKL S2 cells by treatment with Pex1 dsRNA but not by treatment with Dredd dsRNA or by mock treatment (Fig. 3A). Pex1 protein was reduced in lysates of S2 cells treated with Pex1 dsRNA but not in lysates of mock-treated S2 cells or S2 cells treated with Pex7 dsRNA (Fig. 3B). The results of our RNAi analysis in S2 cells demonstrate that loss of function in the majority (13 of 14) of putative Pex genes identified in silico produces abnormal peroxisome assembly phenotypes in Drosophila S2 cells.
Mutation of Pex1 leads to tissue-specific defects in developing embryos PEX1 genes in organisms from yeast to human encode members of the family of AAA-ATPases and are essential for peroxisome assembly. Mutation of human PEX1 results in ZS and is the most common cause of PBD, accounting for 70% or more of all patients (Reuber et al., 1997; Steinberg et al., 2006). Because of the prevalence of the PEX1 mutation in the PBDs, studies of Pex1 mutations were a natural starting point for evaluation of a Drosophila model for the systemic effects of the PBDs and, more specifically, of ZS.
662

Disease Models & Mechanisms DMM

Fig. 3. Detection of Pex1 transcripts by RT-PCR and of Pex1 protein by immunoblotting. (A)The specific bands corresponding to the expression levels of Pex1 and the gene Rpl32, encoding a ubiquitously expressed ribosomal protein and used as a loading control, are indicated by arrows. Heterozygous l(3)70Da animals are indicated by /+. The mRNA isolated from l(3)70DaS4648 homozygotes produced no Pex1-specific band, whereas l(3)70Da1 homozygous animals showed a severe reduction in this band compared with the Rpl32 loading control. Similarly, the level of Pex1 transcript was reduced in the dsRNA-treated S2 cells compared with untreated, mock-treated or S2 cells treated with a dsRNA that targets the Dredd gene, which is involved in the immune response. For each set of primers, specific amounts of the wild-type RT-reaction were analyzed to confirm that 2.5l of experimental sample yielded a product within the linear range of amplification by the subsequent PCR. (B)Pex1 protein is reduced specifically in cells treated with dsRNA to the Pex1 transcript. Lysates of untreated S2 cells, mock-treated S2 cells, S2 cells treated with dsRNA to Pex1 mRNA and S2 cells treated with dsRNA to Pex7 mRNA were separated by SDS-PAGE and subjected to immunoblotting with anti-Pex1 protein antibodies. Reduced levels of Pex1 protein are observed only in the lane containing lysate of S2 cells treated with dsRNA to Pex1 mRNA. A protein detected nonspecifically by the antibodies to Pex1 protein serves as a control for protein loading. Numbers at left represent the migrations of molecular mass standards in kDa.

To study the functional consequences of compromised peroxisome biogenesis in the early development of Drosophila, we performed an in-depth phenotypic characterization of how lossof-function mutations in the Pex1 gene affect basic tissue patterning, survival and the functional response of the organism in terms of altered gene expression. Both P-element and X-rayinduced mutations in Pex1, l(3)70Das4868 and l(3)70Da1, respectively, were characterized in terms of defects during early development. These alleles are lethal when homozygous, in trans or when heterozygous to a large deletion, Df(3L)fz-GS1a, that removes the chromosomal region 70D2-70E5, which includes the Pex1 locus. However, we noted that the genomic region encoding
dmm.biologists.org

Drosophila peroxisome assembly

RESEARCH ARTICLE

Disease Models & Mechanisms DMM

the 3-UTR of the Pex1, l(3)70Da1 mRNA overlaps the genomic region encoding the 5-UTR of the adjacent gene, breathless (btl) (Glazer and Shilo, 1991; Adams et al., 2000). The primary phenotype associated with mutation of btl is defective tracheal development. No tracheal defects were observed in homozygous l(3)70Das4868 (Pex1s4868) or l(3)70Da1 mutants (data not shown). However, to avoid the possibility of confounding phenotypes, we used only the alleles l(3)70Das4868 and l(3)70Da1, which harbor smaller mutations affecting only the 5 region of the Pex1 gene, for subsequent developmental characterizations. Finally, ubiquitous expression of a UASp-Pex1 transgene via Tub-GAL4 in the early embryo allowed homozygous l(3)70Das4868 and l(3)70Da1 mutants to survive past the second larval instar, confirming that the early larval lethality observed in these mutants was caused by loss of Pex1. RT-PCR and immunoblotting confirmed specific reduction in the Pex1 transcript and protein expression in homozygous mutant l(3)70Das4868 and l(3)70Da1 larvae 2 days after hatching (Fig. 3A). Heterozygous Pex1s4868 flies grew (Fig. 4) and survived (Fig. 5) like wild-type flies and, within 2 hours of hatching, both wild-type and heterozygous Pex1 larvae exhibited the same coordinated locomotory activity in their movement towards a food source (Fig. 6). This is in contrast to animals homozygous for the l(3)70Das4868 allele, which grew much more slowly than wild-type or heterozygous flies (Fig. 4) and died at the L1 or L2 stage (Fig. 5). These homozygous mutant larvae were consistently much smaller than wild-type or heterozygous larvae, exhibited little or no coordinated locomotion when placed on agar plates, failed to show effective feeding and exhibited developmental delay (Fig. 6). There is some phenotypic pleiotropy associated with the l(3)70Das4868 and l(3)70Da1 alleles, and, in extreme cases, homozygous mutants were unable to fully crawl out of their eggshells and died within a few hours of hatching. Overall, our results demonstrate that Pex1, like PEX1 for humans, is essential for normal embryonic development.

Fig. 5. Pex1s4868 homozygous flies have a reduced lifespan. Survival curve of Pex1s4868 homozygous (green), Pex1s4868 heterozygous (red) and wild-type (blue) flies. All Pex1s4868 homozygous flies died by day 6 (pupariation). Wildtype and heterozygous larvae pupated on day 6. Error bars indicate standard deviation.

Developing Pex1 homozygous mutant embryos do not show obvious muscle defects The locomotory defects observed in Pex1s4868 homozygous larvae (Fig. 6) could be attributed to aberrations in either neuronal or muscle development (or both). We therefore assayed the developing musculature in Pex1s4868 mutant larvae to determine whether the absence of Pex1 affects muscle development in terms of gross morphology. Staining with an antibody to myosin (MAC147) showed an essentially wild-type pattern of musculature in the homozygous mutant embryos: both wild-type and homozygous mutant late (stage 15) embryos exhibited an evenly repeated pattern of longitudinal and oblique muscles (Fig. 7). Pex1 homozygous mutants exhibit malformed central and peripheral nervous systems Patients affected with a PBD present symptoms arising from significant defects in the CNS and peripheral nervous system (PNS), including mental retardation, seizures, muscular hypotonia and absence of deep tendon reflexes (Steinberg et al., 2006). We

Fig. 4. Pex1s4868 homozygous larvae exhibit defects in growth. Each data point represents the average size in mm2 of 20 randomly selected Pex1s4868 homozygous (green), Pex1s4868 heterozygous (purple) and wild-type (blue) larvae. On day 4, the mean area, which represents growth, of Pex1s4868 homozygous larvae is significantly reduced, as compared with wild-type and heterozygous larvae (P<0.0001). Wild-type and heterozygous larvae did not show any statistically significant difference in growth. Error bars indicate standard deviation.
Disease Models & Mechanisms

Fig. 6. Pex1s4868 homozygous larvae are smaller in size than wild-type and heterozygous larvae, and fail to show coordinated movement towards food. (A)A histogram reporting the percentage of larvae that reach food in a fixed period of 20 minutes. All wild-type and heterozygous larvae, but no homozygous larvae, were able to reach the food source in the prescribed time. (B)Images of 5-day-old wild-type, Pex1s4868 heterozygous and Pex1s4868 homozygous larvae. Homozygous larvae are much smaller than heterozygous or wild-type larvae.

663

RESEARCH ARTICLE

Drosophila peroxisome assembly

Fig. 7. Pex1s4868 homozygous embryos exhibit an essentially wild-type musculature. Wild-type and Pex1s4868 homozygous mutant embryos (stage 15) were analyzed by immunofluorescence microscopy using monoclonal antibody MAC147 to myosin, which recognizes all muscle. Anterior is at right in all images. Lateral views are shown, dorsal is up. Scale bar: 100m.

therefore examined the overall organization of the CNS and PNS in embryos homozygous for Pex1s4868. Stage 15 embryos were stained with antibodies that decorate specifically the CNS and PNS, and were observed by immunofluorescence microscopy (Fig. 8). Monoclonal antibody BP102 stains axons of the CNS (anti-CNS axons) and is an excellent marker for the pattern of commissures and connectives in the CNS of embryos. Wild-type embryos decorated with BP102 showed a ventral nerve cord (VNC) with well-organized and well-formed anterior and posterior commissures and longitudinal connectives. By contrast, Pex1s4868 homozygous embryos showed malformation of the VNC, with a lack of some commissures, breaks in the longitudinal connectives and the presence of underdeveloped commissures, resulting overall in a widening of the distance between the longitudinal connectives of the VNC (Fig. 8). Monoclonal antibody BP104 (anti-Neuroglian; Nrg) stains all neurons of the CNS and PNS and a small number of non-neuronal support cells in the PNS. We observed profound differences in staining between wild-type and homozygous mutant embryos when using anti-Nrg antibody (Fig. 8). In contrast to the CNS and PNS of wild-type embryos, Pex1s4868 homozygous embryos exhibited a marked loss and hypoplasia of neurons in the PNS, evidence of neuronal degeneration, disorganization of the neuronal pattern in the CNS and PNS, and severely malformed developing eye discs. Staining with monoclonal antibody 22C10 (anti-Futsch) also showed dramatic differences in the CNS and PNS of wild-type and homozygous mutant embryos (Fig. 8). Wild-type embryos again exhibited a well-organized PNS and distinct neuron and/or neurite subsets within the VNC, whereas homozygous mutant embryos exhibited severe disruption, disorganization and loss of both PNS neurons and VNC neuron and/or neurite subsets.
Loss of Pex1 causes disorganization in specific subsets of CNS and PNS neurons We used monoclonal antibodies to Even-skipped (Eve; 2B8) and Fasciclin-2 (Fas2; 1D4) to probe more deeply the organization of the CNS and PNS in wild-type and Pex1s4868 homozygous mutant embryos (Fig. 9). Antibody to Eve stains the nuclei of a small subset of neurons in the CNS, whereas anti-Fas2 antibody recognizes the surface of a subset of neurons and axons in the VNC, including some motor neuron axons that innervate striated muscle cells in the periphery of the embryo. Eve expression in the CNS of a wildtype embryo is well organized and well defined when compared with that of the homozygous mutant embryo. However, the pattern
664

Disease Models & Mechanisms DMM

Fig. 8. The overall pattern of CNS and PNS development is abnormal in Pex1s4868 homozygous embryos. Wild-type and Pex1s4868 homozygous embryos (stage 15) were analyzed by immunofluorescence microscopy using monoclonal antibodies BP102 (anti-CNS axons), BP104 (anti-Nrg; recognizing CNS and PNS neurons) and 22C10 (anti-Futsch; recognizing neuron and axon subsets of the CNS and PNS). Anterior is to the right in all images. In lateral views, dorsal is up for BP102 and 22C10 and down for BP104. ac, anterior commissure; lc, longitudinal connective; pc, posterior commissure; VNC, ventral nerve cord. Scale bar: 100m.

of Eve expression in the anal plate is well organized in both wildtype and mutant embryos. Staining of homozygous embryos with anti-Fas2 antibody showed severe abnormalities in the VNC, extensive hypoplasia of developing eye discs and, in the brain, axon mislocalization and a reduction in the number of motor neurons and abnormalities in their organization (Fig. 9). Anti-Cut antibody is known to demark a subset of cells in the PNS, specifically the nuclei of cells of external sensory organ precursors, including the Malpighian tubules, which are the Drosophila counterpart of the mammalian kidney. Pex1s4868 homozygous embryos exhibited massive abnormalities in the structure of developing Malpighian tubules and in the anterior and posterior spiracles, together with a severe loss of neurons in the CNS (Fig. 9).
dmm.biologists.org

Drosophila peroxisome assembly

RESEARCH ARTICLE

Disease Models & Mechanisms DMM

Fig. 10. Glial cells are disorganized in Pex1s4868 homozygous embryos. Wild-type and Pex1s4868 homozygous embryos (stage 15) were analyzed by immunofluorescence microscopy using monoclonal antibodies 8D12 (antiRepo; recognizing all glial cells except midline glia) and 10D3 (anti-Wrapper; recognizing midline glia). Anterior is to the right in all images. In lateral views, dorsal is up. Scale bar: 100m.

Fig. 9. CNS and PNS neurons are disorganized in Pex1s4868 homozygous embryos. Wild-type and Pex1s4868 homozygous embryos (stage 15) were analyzed by immunofluorescence microscopy using monoclonal antibodies 2B8 (anti-Eve; recognizing the nuclei of a subset of CNS neurons), 1D4 (antiFas2; recognizing motor neurons and their axons in the ventral nerve cord) and 2B10 (anti-Cut; recognizing the nuclei of cells of external sensory organ precursors). Anterior is to the right in all images. In lateral views, dorsal is down for anti-Eve, and up for Fas2 and anti-Cut. ap, anal plate; asp, anterior spiracle; mt, Malpighian tubules; psp, posterior spiracle. Scale bar: 100m.

Pex1 homozygous mutant embryos also show disorganized glia Demyelination of axons occurs in the CNS of PBD patients (Steinberg et al., 2006). Although Drosophila lack myelin, glial cells perform a similar function in flies and humans (Freeman and Doherty, 2006). We used monoclonal antibodies to Reversed polarity (Repo; 8D12) and Wrapper (10D3) to label the glial cells of embryos (Fig. 10). Repo is a ubiquitous glial marker found in the nuclei of all glial subtypes and many PNS glia-support cells, but it is not expressed in the midline glia that ensheathe commissural axons. We used anti-Wrapper antibody to recognize midline glia. A dramatic disorganization of glia was observed in Pex1s4868 homozygous embryos compared with wild-type embryos. Taken altogether, our data demonstrate an essential role for the Pex1 gene in the development of the fly nervous system.
Disease Models & Mechanisms

Tissue defects correlate with genome-wide changes in gene expression in Pex1 mutant embryos Because so little is known regarding how mutations in the different PEX genes lead to the underlying pathology of PBD, we reasoned that an unbiased systems level approach would be a good starting point for elucidating new pathways and mechanisms. One strategy of potential use is a comprehensive expression analysis of transcription using microarrays. Loss of peroxisome function is expected to result in the dysregulation of metabolic pathways that could potentially lead to pan-genomic changes in gene expression at each of the cellular, tissue and organismal levels. We performed a comparative genome-wide RNA expression analysis of transcription between wild-type and Pex1s4868 homozygous embryos (Fig. 11). Microarray analysis showed that the expression of 551 distinct protein coding genes showed a consistently greater than threefold difference in expression in 48hour-old larvae homozygous for the Pex1s4868 mutation, compared with wild-type larvae (Fig. 11B). Of these genes, 396 were upregulated (supplementary material Tables S2, S4) and 145 were downregulated (supplementary material Tables S3, S5) in their expression vis--vis the wild-type condition. Notably, genes that were upregulated could be clustered by their predicted Gene Ontology (GO) annotations into several groups whose protein functions correlate strongly with the observed phenotypes of the Pex1 homozygous mutant (Fig. 11C,D). These functions include neural development and activity; behavior; respiration; nucleotide biosynthesis; peptidase function, and adenylate cyclase and GTPase activities; and cell death. Additional pathways whose gene expression was upregulated include those involved with innate immunity, metabolism of purines and mitotic regulation (Fig. 11C; supplementary material Table S2). Genes whose expression was strongly downregulated in the Pex1 homozygous mutant included, as expected, Pex1 itself, as well as genes coding for transcription
665

RESEARCH ARTICLE

Drosophila peroxisome assembly

Fig. 11. Systems level view of changes in gene expression in Pex1s4868 homozygous embryos. (A)Relative positive and negative changes in gene expression from an Affymetrix GeneChip arrays performed in triplicate using wild-type and Pex1s4868 homozygous embryos. (B)Heat maps for genes with more than a threefold change in expression levels in Pex1s4868 homozygous embryos. Bright red indicates a greater than tenfold decrease in expression levels, whereas bright green indicates a greater than tenfold increase in expression levels for genes annotated in FlyBase (http://www.flybase.org). The three color columns each represent the results from an individual experiment. (See supplementary material Table S4 for the corresponding list of genes for each row in the heat map.) (C,D) Genes whose expression levels were increased greater than threefold (C) or decreased more than threefold (D) in Pex1s4868 homozygous embryos as compared with wild-type are grouped according to gene ontology using the ClueGO plugin (http://www.ici.upmc.fr/cluego/cluegoDownload.shtml) for Cytoscape (http://www.cytoscape.org).

Disease Models & Mechanisms DMM

666

dmm.biologists.org

Drosophila peroxisome assembly

RESEARCH ARTICLE

factors involved in gametogenesis and proteins involved in taste sensation, proteolysis and vesicular transport (Fig. 11D; supplementary material Table S3). To control for a potential bias of association based on GO annotation, we also clustered randomly generated lists of genes, which typically generated fewer and sparser networks around annotations distinct from those generated by our microarray data (data not shown). However, nucleus organization was a common annotation that appeared in both the randomly generated lists and our list, which might indicate that this annotation is an over-represented GO term.
DISCUSSION Peroxisome assembly in Drosophila is mechanistically similar to human peroxisome assembly To make robust conclusions regarding the suitability of Drosophila as a model for PBDs, compared with other models, it was necessary to determine whether the overall gene pathways required for peroxisome biogenesis in Drosophila are largely conserved with those in humans. Our systematic RNAi knockdown of 14 Drosophila genes that are homologous to mammalian and/or yeast PEX genes in S2 cells confirmed that at least 13 are bona fide Pex genes. Homologs of these genes have been shown to be required for peroxisome assembly in organisms from yeast to human (reviewed in Platta and Erdmann, 2007; Schrader and Fahimi, 2008). Importantly, mutation in 12 of these genes causes PBD and makes up 12 of the 13 known complementation groups in the PBDs. The thirteenth complementation group belongs to mutations in Pex26, a gene that we and others were unable to find in Drosophila (Schlter et al., 2006). Pex26 is the docking site for the AAAATPases Pex1 and Pex6 in humans, which are necessary for the recycling of the matrix protein receptor Pex5 (Matsumoto et al., 2003). However, this receptor is not conserved across eukaryotes, and other unrelated receptors have been characterized in yeast (Elgersma et al., 1997; Birschmann et al., 2003) and plants (Goto et al., 2011). Despite being unrelated, all three of these proteins function similarly. Therefore, Drosophila most probably possesses a receptor for Pex1 and Pex6 that remains to be discovered. Cells from PBD patients typically demonstrate increased cytosolic localization of peroxisomal matrix proteins. RNAi of Drosophila Pex2, Pex6, Pex13, Pex16 or Pex19 dsRNA produced clear reductions in the amounts of the corresponding mRNAs in treated S2 cells (supplementary material Fig. S1), whereas RNAi of the other Pex genes produced a less robust, yet still demonstrable, reduction in the corresponding mRNAs. Despite this variation in the reduction of mRNA levels, knockdown of Pex1, Pex2, Pex3, Pex6, Pex12, Pex13, Pex14 or Pex16 reduced or eliminated the punctate structures that are characteristic of peroxisomes, and mislocalized a GFP-SKL reporter to the cytosol. Although the knockdown of Pex5 by RNAi was somewhat variable, cells consistently showed an exclusively cytosolic signal for the GFPSKL reporter. Pex5 thus probably encodes the PTS1 receptor in Drosophila. Conversely, Pex7 functions as the receptor for proteins targeted to peroxisomes by PTS2, independently of the PTS1(SKL) targeting sequence. This makes the knockdown of Pex7 somewhat difficult to interpret, because we would not expect a change in the characteristic punctate pattern produced by GFP-SKL. Likewise, the presence of a putative PEX20 homolog in Drosophila would suggest that there are parallels in the peroxisomal protein-import pathways
Disease Models & Mechanisms

of insects and fungi (Titorenko et al., 1998). Several species of yeast have been shown to rely on the interaction between Pex7p and a Pex20-like protein for PTS2 protein import, whereas mammalian cells import PTS2 proteins via an interaction between Pex7 and a larger splice variant of Pex5 known as Pex5L (Girzalsky et al., 2010). Pex19 protein has been shown to be required for peroxisomal membrane formation and has been suggested to function as the receptor for peroxisomal membrane proteins and/or as a chaperone acting to stabilize membrane proteins at the peroxisomal membrane. Absence of Pex19 in most organisms leads to an inability of cells to assemble peroxisomes and an absence of identifiable peroxisomal membrane structures (Hettema et al., 2000). Interestingly, although we could show a clear and reproducible reduction in the mRNA of the putative Drosophila homolog of PEX19 in RNAi-treated S2 cells (supplementary material Fig. S1), this nevertheless did not prevent the assembly of peroxisomes but instead led to a reduction in the number and increase in the size of punctate structures containing GFP-SKL and partial mislocalization of GFP-SKL to the cytosol. These differences in peroxisomal phenotype between the cells of other organisms lacking Pex19 protein and S2 cells subjected to RNAi for Pex19 mRNA could arise for a variety of reasons: (1) Pex19 might not be the true Drosophila homolog of PEX19 in other organisms. However, the degree of sequence similarity between human and Drosophila Pex19 (Fig. 1) makes this unlikely. (2) Pex19 protein might have functions in peroxisome biogenesis that are different from those of Pex19 protein in other organisms. Because of the predicted multivariate nature of functions attributed to Pex19, further study of Pex19 might provide a better understanding of Pex19 function. (3) Drosophila might have another protein whose function in peroxisome biogenesis and peroxisomal membrane formation is redundant to that of Pex19 protein. The de novo synthesis of peroxisomes from the endoplasmic reticulum (ER) has been shown to rely on additional proteins, such as Dsl1p, Sec39p and Sec20p, in Saccharomyces cerevisiae (Perry et al., 2009), and essential secretory proteins might play a more prominent role in this process in Drosophila. Other proteins, such as the p24 family of proteins (Marelli et al., 2004; Otzen et al., 2006), might also affect this trafficking pathway. The Drosophila system remains an excellent model organism for the discovery of new secretory mechanisms (Bard et al., 2006). (4) RNAi reduction of Pex19 was insufficient to give a loss-of-function phenotype. A similar phenotype has been described for the yeast Yarrowia lipolytica, in which deletion of the PEX19 gene results in the accumulation of peroxisomal structures that are similar in morphology to wild-type peroxisomes but are defective in matrix protein import (Lambkin and Rachubinski, 2001). Together, these observations suggest that Pex19 might participate in several aspects of peroxisome biogenesis and be tailored for one or more aspects in individual organisms (Mast et al., 2010; Ma et al., 2011). PEX11 is conserved among eukaryotes, with the role of controlling peroxisome division and proliferation (Fagarasunu et al., 2007). A loss of Pex11 protein produces cells with decreased numbers of enlarged peroxisomes (Erdmann and Blobel, 1995). Drosophila Pex11 probably functions similarly to PEX11 genes in other organisms, because RNAi to PEX11 transcript resulted in S2 cells containing few and enlarged peroxisomes but were unaffected in their ability to import GFP-SKL into peroxisomes.
667

Disease Models & Mechanisms DMM

RESEARCH ARTICLE

Drosophila peroxisome assembly

Drosophila Pex23 exhibits limited similarity to Y. lipolytica Pex23p. Nevertheless, the peroxisome phenotype of RNAi-treated S2 cells (Fig. 2) is consistent with a role for Drosophila Pex23 in peroxisome biogenesis. Interestingly, the region of highest similarity between Drosophila Pex23 and Y. lipolytica Pex23p is in the unique dysferlin motif that is common to all members of the Pex23 protein family. Dysferlin motifs are thought to aid proteins in binding to certain phosphoinositides such as phosphatidylinositol 4-phosphate. This feature is proposed to aid the Pex23 family of proteins in their function of controlling peroxisome number and size in different yeast species (Brown et al., 2000; Vizeacoumar et al., 2004; Yan et al., 2008). S2 cells treated by RNAi against the transcript of Pex23 had increased numbers of small peroxisomes as compared with control S2 cells, and no mislocalization of GFP-SKL to the cytosol. Therefore, notwithstanding the evolutionary distance between flies and yeasts, Drosophila has apparently retained some PEX genes that have heretofore only been reported in yeasts and have maintained their involvement in peroxisome biogenesis.
Pex1 mutants mirror the early developmental defects associated with human PBDs A recent study of flies harboring mutant Pex2 and Pex10 peroxins has demonstrated a requirement for peroxisomes in normal spermatogenesis and the control of the levels of very-long-chain fatty acids in adult Drosophila (Chen et al., 2010). However, it is difficult to ascertain whether gametogenesis is affected in PBD patients, because mutation of the 13 PEX genes involved in PBDs cause early developmental defects and usually death within 1 year (Shimozawa et al., 2005; Steinberg et al., 2006). Because inherited mutations of PEX1 are the most common cause of ZS, and of PBDs in general (Reuber et al., 1997; Steinberg et al., 2006), we wanted to examine the effect of Drosophila Pex1 mutations on early development. The organismal phenotype associated with loss of Pex1 protein is strikingly similar to the ZS phenotypic spectrum. Pex1 mutant larvae were smaller in size than their wild-type counterparts, failed to show coordinated movement or effective feeding habits, and exhibited severe malformations of the nervous system, with loss and/or mislocalization of axons and neurons in both the CNS and PNS during embryonic development. There are additional parallels between the abnormalities exhibited by ZS patients and flies mutant in the Pex1 gene. ZS patients have severely demyelinated axons in both the CNS and PNS (Powers et al., 1985; Steinberg et al., 2006). Although Drosophila axons are not myelinated, Pex1 mutant flies have extensively disorganized glial cells, which provide support and insulation to the axons of the CNS and PNS of flies in a manner akin to myelination in mammals (Freeman and Doherty, 2006). Pex1 mutants also exhibit malformed Malpighian tubules, the functions of which are similar to those of mammalian kidneys. ZS patients also show structural abnormalities in the kidney (Steinberg et al., 2006). The use of a simple genetic model such as Drosophila allows the examination of the cellular and organismal response to loss of Pex1 in terms of gene expression. Thus, for the first time, we were able to evaluate the results of the loss of Pex1 at a systems level using transcriptomic profiling. In mutant animals, the primary systemic molecular response to loss of peroxisome function can be linked to changes in 551 of the 13,767 protein coding genes (approximately 4%). Furthermore, the majority of these genes can be functionally
668

grouped into relatively few genetic pathways. However, the diversity of genetic pathways altered by loss of peroxisome function underscores the widespread and varied responses seen in both the clinical and biochemical presentation of PBDs. Our transcriptomic data reveal large-scale alterations in the expression of genes involved in chitin metabolism and puparial adhesion. This might implicate peroxisomes in the metabolic pathways of complex polysaccharides such as chitin, or might reflect changes in the secretory pathway through dysregulation and loss of peroxisome function. It has recently been demonstrated in the yeast Pichia pastoris that non-conventional secretion of acyl-CoA binding protein requires proper peroxisome metabolism (Manjithaya et al., 2010). These findings underscore our present lack of knowledge regarding peroxisome biology, which is essential for better understanding the disease progression of PBDs. The increase in expression of genes involved in innate immunity and humoral responses might be due to a response to increased levels of unused peroxisomal metabolites. Plasma levels of the plasmalogen precursors di- and trihydroxycholestanoic acid are increased in PBD patients, as is a C29-dicarboxylic acid that is not typically found in the serum of normal individuals (Ferdinandusse et al., 2009). The accumulation of these compounds and long-chain fatty acids that are esterified to cholesterol (Kovacs et al., 2004; Kovacs et al., 2009) are thought to contribute to the pleiotropic defects exhibited by PBD patients (Mast et al., 2010). However, changes in the expression of genes that are involved in innate immunity might underlie a currently unexplored role for peroxisomes in immune responses. Recently, mammalian peroxisomes were shown to function as signaling platforms for antiviral innate immunity (Dixit et al., 2010). The upregulation of genes involved in the immune response in Pex1 mutants suggests that peroxisomes might also function in modulating the immune response of Drosophila. Our profiling of changes in gene expression in response to loss of peroxisome function provides a tractable method to monitor disease progression. It might also prove to be a suitable tool for screening novel therapeutic drug candidates and monitoring disease regression in response to novel therapies of PBDs. Current attempts at finding small molecules and diet regimens that might alleviate symptoms of PBDs require extensive and time-consuming biochemical verification (Zhang et al., 2010; Dranchak et al., 2010). This process could be aided by transcriptomic profiling as a way to identify treatments that return gene expression profiles back to normal levels. On the basis of the multiple classes of defects observed in our RNAi analysis in S2 cells, we have confirmed that at least 13 of the 14 known Drosophila Pex genes are required for what are probably different aspects of peroxisome assembly. Moreover, we have shown that specific developmental abnormalities in the developing embryonic nervous system occur when the Pex1 gene is mutated. Taken together, our findings make a compelling argument for the use of Drosophila as a valid and tractable model system with which to investigate the roles of peroxisomes and the effects of PBD on the growth, development and response to the environment of a multicellular organism.
METHODS Cell culture Drosophila S2 cells were grown at 25C in Schneiders Drosophila Medium (Invitrogen) supplemented with 10% heat-inactivated
dmm.biologists.org

Disease Models & Mechanisms DMM

Drosophila peroxisome assembly

RESEARCH ARTICLE
Day 2: 2105 cells in 500 l of SFX culture medium were added to each well of a 24-well dish (BD Biosciences); 20 g of dsRNA to a gene of interest were added to one well, followed by gentle horizontal mixing; plates were sealed with Parafilm M and incubated at 25C for 96 hours (days 2-5). Day 6: the cells in each well were split 1:2, and an additional 20 g of dsRNA were added; the plates were resealed with Parafilm M and incubated at 25C for 48 hours (days 6-7). Day 8: 10% of the remaining volume of cells was transferred to LabTekII chambered coverslips (Nalge Nunc) for microscopic examination of peroxisomes; total RNA was extracted from the remaining cells using RNeasy (Qiagen) and quantified with a NanoVue spectrophotometer (GE Healthcare); 200 ng of total RNA were used for reverse transcription using Superscriptase II (Invitrogen) and oligo-dT as the anchor primer; 2 l of the resulting cDNA and gene-specific primers were then used in a 50-l PCR reaction; for semi-quantitative PCR, specific Pex gene primers and control primers for a mRNA encoding a ribosomal protein (RP49) were used; the number of PCR cycles for each set of gene-specific primers was tested to ensure that the resulting reaction was within the linear range of amplification; products were separated by electrophoresis on 1% agarose gels; band intensity was measured using ImageJ and normalized to the control RP49 signal.
Preparation of antibodies to Pex1 protein and immunoblotting Antibodies that recognize Pex1 were raised against a fusion of maltose-binding protein and the N-terminal 200 amino acids of Pex1 in guinea-pig and tested for specificity by immunoblotting of an S2 cell lysate using enhanced chemiluminescence (GE Healthcare). The protocol for raising the antibody is essentially as described before (Eitzen et al., 1997). Drosophila stocks Oregon R (wild-type) and Pex1 (CG6760) strains l(3)70Das4868 and l(3)70Da1 were obtained from the Bloomington Drosophila Stock Center at Indiana University. Stocks were maintained on standard Bloomington medium at 25C, and embryos were collected at 25C on apple juice agar plates. To facilitate the identification of homozygous mutant embryos, lethal Pex1 mutations were maintained over the w; Sb1/TM3, P{ActGFP}JMR2, Ser1 balancer. UASp-Pex1 was made by TOPO cloning of a PCR fragment corresponding to the open reading frame of Pex1 into pENTR/D (Invitrogen). This was transferred using Clonase LR (Invitrogen) into the plasmid pPW (Terence Murphy), which was transformed into w1118 embryos by P-element recombination. Larval survival, growth and mobility assays Parallel cultures of Oregon R and l(3)70Das4868 adult flies were maintained for 1 week in a standard 2 L population cage at 25C. Flies were allowed to lay eggs for 1 hour on apple juice agar plates, and cohorts of 100 randomly selected embryos of each genotype were placed on fresh apple juice agar plates with yeast paste at the center of the plates. The lifespan of l(3)70Das4868 homozygous mutants was determined by observing larvae daily from hatching until death. Larvae were fed and watered regularly and kept in a 25C incubator. The number of survivors was recorded daily. All homozygous mutants were dead by day 6. However, wild-type and l(3)70Das4868 heterozygous larvae molted normally at day 6 and
669

fetal bovine serum, 50 U penicillin/ml and 50 g streptomycin sulfate/ml (Invitrogen). S2 cells constitutively expressing the fluorescent peroxisomal chimeric protein GFP-SKL (Kural et al., 2005) were kindly provided by Ronald Vale (University of California, San Francisco, CA).
Fluorescence microscopy Embryos were fixed as described (Hughes and Krause, 1999) and stained with the following antibodies at 1:100 dilution (obtained from the Developmental Studies Hybridoma Bank, University of Iowa, IA): mouse anti-CNS axons (BP102), mouse anti-Neuroglian (BP104), mouse anti-Fas2 (1D4), mouse anti-Repo (8D12) and mouse antiWrapper (10D3) (all developed by Corey Goodman, University of California, Berkeley, CA); mouse anti-Futsch (22C10; developed by Seymour Benzer, California Institute of Technology, Pasadena, CA); mouse anti-Even-skipped (2B8; developed by Kai Zinn, California Institute of Technology, Pasadena, CA); and mouse anti-Cut (2B10; developed by Gerald M. Rubin, University of California, Berkeley, CA). Rabbit anti-GFP (Invitrogen) and rat anti-myosin (Abcam) were used at 1:1000 dilution. Alexa-Fluor-488-, Alexa-Fluor-568-, Cy2- and Cy5-conjugated secondary antibodies (Invitrogen, Jackson ImmunoResearch) were used at 1:2000 dilution to recognize primary antibodies. To reduce the effects of spherical aberrations, embryos were optically cleared in methyl salicylate after preparation for immunofluorescence (MacDonald and Rubel, 2008). Images were obtained using a UPlanFl 20/0.5 NA air objective on an IX81 inverted epifluorescence microscope (Olympus) equipped with a CoolSNAP HQ digital camera (Roper Scientific) and an X-Cite 120 PC fluorescent illumination system (EXFO Life Sciences). Images of RNAi-treated S2 cells were acquired using a PlanApochromat 63/1.4 NA oil DIC objective on an Axiovert 200 inverted microscope equipped with a LSM510 META confocal scanner (Carl Zeiss). GFP-SKL was excited with a 488-nm argon laser and its emission collected with a 505-nm long-pass filter. Images were deconvolved using algorithms provided by Huygens Professional Software (Scientific Volume Imaging BV, The Netherlands). For deconvolution, 3D data sets were processed to remove noise and reassign blur by an iterative classic maximum likelihood estimation widefield algorithm and a confocal algorithm, respectively, and an experimentally derived point spread function. Transmission images of S2 cells were processed by applying a Gaussian filter in Huygens, and blue color was applied to the transmission images using Imaris software (Bitplane). Transmission images were processed to maximize the fluorescent signal while maintaining cell outlines. Imaris was then used to render the deconvolved 3D data set to the processed transmission image. Final figure assembly was performed in Adobe Photoshop, Adobe Illustrator or Adobe InDesign. RNAi knockdown of Pex homologs and semi-quantitative PCR Templates for dsRNAs targeting predicted Pex homologs were isolated from a template library (Foley and OFarrell, 2004). A control template to make dsRNA that specifically targets Dredd was a gift from Edan Foley, University of Alberta, Canada. dsRNAs were synthesized using the T7 RiboMAX kit (Promega) using template-specific primers (Foley and OFarrell, 2004). The dsRNA treatment of cells was performed as follows: day 1: cells were split 1:2 with fresh SFX culture medium (Invitrogen).
Disease Models & Mechanisms

Disease Models & Mechanisms DMM

RESEARCH ARTICLE

Drosophila peroxisome assembly

TRANSLATIONAL IMPACT
Clinical issue
Mutation of genes that are responsible for the biogenesis and maintenance of peroxisomes cause a spectrum of debilitating disorders known as peroxisome biogenesis disorders (PBDs). Individuals with PBDs exhibit an array of clinical characteristics including ichthyosis, facial dysmorphology, retinitis pigmentosa, cataracts, deafness, hepatic disease, adrenal insufficiency, renal cysts, chondrodysplasia punctata, hypotonia, cerebral dysgenesis, and various forms of developmental, mental and physical retardation. Variability in the presentation of defects across the spectrum of PBDs reflects the central role of peroxisome metabolism in human development. The underlying cause of pathophysiology of the PBDs remains poorly understood and would benefit from a disease model that recapitulates aspects of the patient phenotype, enables functional characterization of disease mechanisms and increases understanding of the contribution of peroxisomes to normal physiology.

cross-sectional area was measured using ImageJ software. Plates were monitored until day 4. The growth assay was repeated three times with more than 20 larvae in each experimental group. To assay larval mobility, larvae of each genotype were placed at one end of an apple juice agar plate and allowed to crawl to yeast paste placed at the other end of the plate. The distance crawled in 10 minutes was measured. 5-day-old larvae of each genotype were digitally photographed with a SZX12 microscope as above. Images were assembled with ImageJ and Adobe Photoshop.
Microarray analysis Heterozygous l(3)70Das4868/TM3, P{ActGFP}JMR2, Ser1 parents were allowed to lay embryos on apple juice agar plates for 30 minutes at 25C. The plates were then supplemented with yeast paste, and the embryos were allowed to develop for 48 hours at 25C and 70% humidity. Hatched larvae were then hand sorted under a SX61 fluorescence stereomicroscope (Olympus) to select homozygous l(3)70Das4868 larvae, i.e. those lacking GFP. Care was taken to select only larvae judged clearly to be alive as evidenced by their movement. Corresponding wild-type siblings were isolated from w1118/TM3, P{ActGFP}JMR2, Ser1 parents. Total mRNA from 200 mg samples was isolated using Trizol (Invitrogen) and purified using RNeasy. The resulting RNA was tested for quality with an Agilent 2100 Bioanalyzer, converted to labeled complementary RNA (cRNA) and hybridized to Drosophila 2 3-UTR microarrays (Affymetrix). These experiments were run in triplicate. The resulting array data were normalized using the MAS5 algorithm and Expression Console software (Affymetrix). Genes showing a reproducible threefold increase or decrease in expression relative to their expression in wild-type siblings in three independent experiments were clustered based on their functional Gene Ontology annotations and known biochemical pathways using the DAVID package (Huang et al., 2009). Clustering of related genes according to their functional Gene Ontology annotations was done using the Cluego plugin (Bindea et al., 2009) for Cytoscape (www.cytoscape.org/).
ACKNOWLEDGEMENTS We thank Ronald Vale for the GFP-SKL S2 cell line and Edan Foley for the DNA templates to produce the dsRNA library. We would also like to thank Richard Poirier, Elena Savidov, Hanna Kroliczak and Dwayne Weber for their technical help, and members of the Rachubinski laboratory for helpful discussions and support. F.D.M. was supported by a Canada Graduate Scholarship from the Canadian Institutes of Health Research (CIHR) and a Faculty of Medicine and Dentistry 75th Anniversary Award. This work was supported by grant 9208 from the CIHR to R.A.R. and grant 84154 from the CIHR to A.J.S. A.J.S. is a Senior Scholar of the Alberta Heritage Foundation for Medical Research. R.A.R. is an International Research Scholar of the Howard Hughes Medical Institute. COMPETING INTERESTS The authors declare that they do not have any competing or financial interests. AUTHOR CONTRIBUTIONS S.C.H., A.J.S. and R.A.R. designed the research, analyzed the data and wrote the paper. F.D.M., J.L. and M.K.V. performed the experiments and analyzed the data. SUPPLEMENTARY MATERIAL Supplementary material for this article is available at http://dmm.biologists.org/lookup/suppl/doi:10.1242/dmm.007419/-/DC1 REFERENCES Adams, M. D., Celniker, S. E., Holt, R. A., Evans, C. A., Gocayne, J. D., Amanatides, P. G., Scherer, S. E., Li, P. W., Hoskins, R. A., Galle, R. F. et al. (2000). The genome sequence of Drosophila melanogaster. Science 287, 2185-2195. Bard, F., Casano, L., Mallabiabarrena, A., Wallace, E., Saito, K., Kitayama, H., Guizzunti, G., Hu, Y., Wendler, F., DasGupta, R. et al. (2006). Functional genomics dmm.biologists.org

Results
In this study, the authors investigate the use of Drosophila melanogaster as a model in which to study the pathophysiology of PBDs. An in silico interrogation of the Drosophila genome identifies 15 putative homologs (Pex genes) of known human and yeast PEX genes. RNAi-mediated knockdown confirms that 13 of these putative Pex genes are required for peroxisome assembly in Drosophila S2 cells. For example, the Pex1 gene encodes an AAAATPase, which, when mutated in humans, causes the most severe PBD, Zellweger syndrome (ZS). The authors show that knockdown of Pex1 in S2 cells results in mislocalization of a peroxisome-targeted chimeric protein consisting of GFP and the peroxisome-targeting signal Ser-Lys-Leu (GFP-SKL) 3 to the cytosol. Furthermore, analysis of flies carrying a mutant form of the Pex1 gene shows that Pex1 is required for normal fly development, as is its homolog required for normal human development. Furthermore, mutant flies exhibit abnormalities characteristic of ZS patients, including developmental delay, poor feeding, severe structural abnormalities in the peripheral and central nervous systems, and early death. Finally, microarray analysis defines several clusters of genes whose expression varies significantly between wild-type and Pex1-mutant larvae, and that implicate peroxisomal function as a requirement for normal neuronal development, innate immunity, lipid and protein metabolism, gamete formation, and meiosis.

Disease Models & Mechanisms DMM

Implications and future directions

These results demonstrate that D. melanogaster is a useful model for studying PBDs. The finding that, as in humans, developmental programs in D. melanogaster require functional peroxisomes is demonstrated by the gross scale malformations noted in the organization of the central and peripheral nervous systems of mutant flies. In addition, these results reveal that dysregulation of genes involved in innate immunity and mitosis has a previously unappreciated role in disease pathology. Approaches to therapy for PBDs have been hampered owing to the rarity of these diseases and because of the lack of a cost-effective disease model. Future studies can make use of Pex-mutant flies to carry out large-scale screens to identify small molecules that might alleviate or reduce the developmental and physiological abnormalities associated with PBDs, which have thus far not been possible. Classes of molecules that suppress PBD phenotypes in flies could be developed as potential new treatments for PBD patients.

hatched into adult flies by day 10. The survival assay was repeated three times with 100 larvae in each experimental group. To assay larval growth, cohorts of 20 larvae of each genotype that had hatched were selected randomly after 24 hours. Staged larvae were placed on fresh apple juice agar plates with yeast paste at the center of the plates, allowed to grow for an additional 72 hours at 25C, then visualized with a DF PLAPO 1.2 PF objective on a SZX12 microscope (Olympus) equipped with a PC1015 camera (Canon) and digitally photographed every 24 hours after hatching. Larval
670

Drosophila peroxisome assembly

RESEARCH ARTICLE

reveals genes involved in protein secretion and Golgi organization. Nature 439, 604607. Beard, M. E. and Holtzman, E. (1987). Peroxisomes in wild-type and rosy mutant Drosophila melanogaster. Proc. Natl. Acad. Sci. USA 84, 7433-7437. Bilen, J. and Bonini, N. M. (2005). Drosophila as a model for human neurodegenerative disease. Annu. Rev. Genet. 39, 153-171. Bindea, G., Mlecnik, B., Hackl, H., Charoentong, P., Tosolini, M., Kirilovsky, A., Fridman, W., Pags, F., Trajanoski, Z. and Galon, J. (2009). ClueGO: a Cytoscape plug-in to decipher functionally grouped gene ontology and pathway annotation networks. Bioinformatics 25, 1090-1093. Birschmann, I., Stroobants, A. K., van den Berg, M., Schfer, A., Rosenkranz, K., Kunau, W. and Tabak, H. F. (2003). Pex15p of Saccharomyces cerevisiae provides a molecular basis for recruitment of the AAA peroxin Pex6p to peroxisomal membranes. Mol. Biol. Cell 14, 2226-2236. Bridges, C. B. and Brehme, K. S. (1944). The mutants of Drosophila melanogaster. Carnegie Inst. Wash. Publ. 552, 252. Brown, T. W., Titorenko, V. I. and Rachubinski, R. A. (2000). Mutants of the Yarrowia lipolytica PEX23 gene encoding an integral peroxisomal membrane protein mislocalize matrix proteins and accumulate vesicles containing peroxisomal matrix and membrane proteins. Mol. Biol. Cell 11, 141-152. Cauchi, R. J. and van den Heuvel, M. (2006). The fly as a model for neurodegenerative diseases: is it worth the jump? Neurodegener. Dis. 3, 338-356. Chen, H., Liu, Z. and Huang, X. (2010). Drosophila models of peroxisomal biogenesis disorder: peroxins are required for spermatogenesis and very long chain fatty acid metabolism. Hum. Mol. Genet. 19, 494-505. Dixit, E., Boulant, S., Zhang, Y., Lee, A. S. Y., Odendall, C., Shum, B., Hacohen, N., Chen, Z. J., Whelan, S. P., Fransen, M. et al. (2010). Peroxisomes are signaling platforms for antiviral innate immunity. Cell 141, 668-681. Dranchak, P. K., Di Pietro, E., Snowden, A., Oesch, N., Braverman, N. E., Steinberg, S. J. and Hacia, J. G. (2010). Nonsense supressor therapies rescue peroxisome lipid metabolism and assembly in cells from patients with specific PEX gene mutations. J. Cell. Biochem. 112, 1250-1258. Eitzen, G. A., Szilard, R. K. and Rachubinski, R. A. (1997). Enlarged peroxisomes are present in oleic acid-grown Yarrowia lipolytica overexpressing the PEX16 gene encoding an intraperoxisomal peripheral membrane peroxin. J. Cell Biol. 137, 12651278. Elgersma, Y., Kwast, L., van den Berg, M., Snyder, W. B., Distel, B., Subramani, S. and Tabak, H. F. (1997). Overexpression of Pex15p, a phosphorylated peroxisomal integral membrane protein required for peroxisome assembly in S. cerevisiae, causes proliferation of the endoplasmic reticulum membrane. EMBO J. 16, 7326-7341. Erdmann, R. and Blobel, G. (1995). Giant peroxisomes in oleic acid-induced Saccharomyces cerevisiae lacking the peroxisomal membrane protein Pmp27. J. Cell Biol. 128, 509-523. Fagarasanu, A., Fagarasanu, M. and Rachubinski, R. A. (2007). Maintaining peroxisome populations: a story of division and inheritance. Annu. Rev. Cell Dev. Biol. 23, 321-344. Ferdinandusse, S., Denis, S., Faust, P. L. and Wanders, R. J. A. (2009). Bile acids: the role of peroxisomes. J. Lipid Res. 50, 2139-2147. Foley, E. and OFarrell, P. H. (2004). Functional dissection of an innate immune response by a genome-wide RNAi screen. PLoS Biol. 2, E203. Freeman, M. R. and Doherty, J. (2006). Glial cell biology in Drosophila and vertebrates. Trends Neurosci. 29, 82-90. Girzalsky, W., Saffian, D. and Erdmann, R. (2010). Peroxisomal protein translocation. Biochim. Biophys. Acta 1803, 724-731. Glassman, E. and Mitchell, H. K. (1958). Mutants of Drosophila melanogaster deficient in xanthine dehydrogenase. Genetics 44, 153-162. Glazer, L. and Shilo, B. Z. (1991). The Drosophila FGF-R homolog is expressed in the embryonic tracheal system and appears to be required for directed tracheal cell extension. Genes Dev. 5, 697-705. Goto, S., Mano, S., Nakamori, C. and Nishimura, M. (2011). Arabidopsis ABERRANT PEROXISOME MORPHOLOGY9 is a peroxin that recruits the PEX1-PEX6 complex to peroxisomes. Plant Cell 23, 1573-1587. Hettema, E. H., Girzalsky, W., van den Berg, M., Erdmann, R. and Distel, B. (2000). Saccharomyces cerevisiae Pex3p and Pex19p are required for proper localization and stability of peroxisomal membrane proteins. EMBO J. 19, 223-233. Huang, D. W., Sherman, B. T. and Lempicki, R. A. (2009). Systematic and integrative analysis of large gene lists using DAVID bioinformatics resources. Nat. Protoc. 4, 4457. Hughes, S. C. and Krause, H. M. (1999). Single and double FISH protocols for Drosophila. Methods Mol. Biol. 122, 93-101. Kovacs, W. J., Shackelford, J. E., Tape, K. N., Richards, M. J., Faust, P. L., Fliesler, S. J. and Krisans, S. K. (2004). Disturbed cholesterol homeostasis in a peroxisomedeficient PEX2 Knockout Mouse Model. Mol. Cell. Biol. 24, 1-13. Disease Models & Mechanisms

Kovacs, W. J., Tape, K. N., Shackelford, J. E., Wikander, T. M., Richards, M. J., Fliesler, S. J., Krisans, S. K. and Faust, P. L. (2009). Peroxisome deficiency causes a complex phenotype because of hepatic SREBP/Insig dysregulation associated with endoplasmic reticulum stress. J. Biol. Chem. 284, 7232-7245. Kural, C., Kim, H., Syed, S., Goshima, G., Gelfand, V. I. and Selvin, P. R. (2005). Kinesin and dynein move a peroxisome in vivo: a tug-of-war or coordinated movement? Science 308, 1469-1472. Lambkin, G. R. and Rachubinski, R. A. (2001). Yarrowia lipolytica cells mutant for the peroxisomal peroxin Pex19p contain structures resembling wild-type peroxisomes. Mol. Biol. Cell 12, 3353-3364. Lazarow, P. B. (2006). The import receptor Pex7p and the PTS2 targeting sequence. Biochim. Biophys. Acta 1763, 1599-1604. Lazarow, P. B. and Fujiki, Y. (1985). Biogenesis of peroxisomes. Annu. Rev. Cell Biol. 1, 489-530. Ma, C., Agrawal, G. and Subramani, S. (2011). Peroxisome assembly: matrix and membrane protein biogenesis. J. Cell Biol. 193, 7-16. MacDonald, G. and Rubel, E. W. (2008). Three-dimensional imaging of the intact mouse cochlea by fluorescent laser scanning confocal microscopy. Hear. Res. 243, 110. Manjithaya, R., Jain, S., Farr, J. and Subramani, S. (2010). A yeast MAPK cascade regulates pexophagy but not other autophagy pathways. J. Cell Biol. 189, 303-310. Marelli, M., Smith, J. J., Jung, S., Yi, E., Nesvizhskii, A. I., Christmas, R. H., Saleem, R. A., Tam, Y. Y. C., Fagarasanu, A., Goodlett, D. R. et al. (2004). Quantitative mass spectrometry reveals a role for the GTPase Rho1p in actin organization on the peroxisome membrane. J. Cell Biol. 167, 1099-1112. Mast, F. D., Fagarasanu, A., Knoblach, B. and Rachubinski, R. A. (2010). Peroxisome biogenesis: something old, something new, something borrowed. Physiology 25, 347-356. Matsumoto, N., Tamura, S. and Fujiki, Y. (2003). The pathogenic peroxin Pex26p recruits the Pex1p-Pex6p AAA ATPase complexes to peroxisomes. Nat. Cell Biol. 5, 454-460. Moser, A. B., Rasmussen, M., Naidu, S., Watkins, P. A., McGuinness, M., Hajra, A. K., Chen, G., Raymond, G., Liu, A., Gordon, D. et al. (1995). Phenotype of patients with peroxisomal disorders subdivided into sixteen complementation groups. J. Pediatr. 127, 13-22. Otzen, M., Krikken, A. M., Ozimek, P. Z., Kurbatova, E., Nagotu, S., Veenhuis, M. and van der Klei, I. (2006). In the yeast Hansenula polymorpha, peroxisome formation from the ER is independent of Pex19p, but involves the function of p24 proteins. FEMS Yeast Res. 6, 1157-1166. Perry, R. J., Mast, F. D. and Rachubinski, R. A. (2009). Endoplasmic reticulumassociated secretory proteins Sec20p, Sec39p, and Dsl1p, are involved in peroxisome biogenesis. Eukaryot. Cell 8, 830-843. Petriv, O. I., Pilgrim, D. B., Rachubinski, R. A. and Titorenko, V. I. (2002). RNA interference of peroxisome-related genes in C. elegans: a new model for human peroxisomal disorders. Physiol. Genomics 10, 79-91. Platta, H. W. and Erdmann, R. (2007). Peroxisomal dynamics. Trends Cell Biol. 17, 474484. Powers, J. M., Moser, H. W., Moser, A. B., Upshur, J. K., Bradford, B. F., Pai, S. G., Kohn, P. H., Frias, J. and Tiffany, C. (1985). Fetal cerebrohepatorenal (Zellweger) syndrome: dysmorphic, radiologic, biochemical, and pathologic findings in four affected fetuses. Hum. Pathol. 16, 610-620. Raymond, F. L. and Tarpey, P. (2006). The genetics of mental retardation. Hum. Mol. Genet. 15, R110-R116. Reiter, L. T., Potocki, L., Chien, S., Gribskov, M. and Bier, E. (2001). A systematic analysis of human disease-associated gene sequences in Drosophila melanogaster. Genome Res. 11, 1114-1125. Reuber, B. E., Germain-Lee, E., Collins, C. S., Morrell, J. C., Ameritunga, R., Moser, H. W., Valle, D. and Gould, S. J. (1997). Mutations in PEX1 are the most common cause of peroxisome biogenesis disorders. Nat. Genet. 17, 445-448. Snchez-Martnez, A., Luo, N., Clemente, P., Adn, C., Hernndez-Sierra, R., Ochoa, P., Fernndez-Moreno, M. A., Kaguni, L. S. and Garesse, R. (2006). Modeling human mitochondrial diseases in flies. Biochim. Biophys. Acta 1757, 1190-1198. Schlter, A., Fourcade, S., Ripp, R., Mandel, J., Poch, O. and Pujol, A. (2006). The evolutionary origin of peroxisomes: an ER-Peroxisome connection. Mol. Biol. Evol. 23, 838-845. Schrader, M. and Fahimi, H. D. (2008). The peroxisome: still a mysterious organelle. Histochem. Cell Biol. 129, 421-440. Shimozawa, N. (2011). Molecular and clinical findings and diagnostic flowchart of peroxisomal diseases. Brain Dev. doi:10.1016/j.braindev.2011.03.004 (Epub ahead of print). Shimozawa, N., Imamura, A., Zhang, Z., Suzuki, Y., Orii, T., Tsukamoto, T., Osumi, T., Fujiki, F., Wanders, R. J. A., Besley, G. et al. (1999). Defective PEX gene products correlate with the protein import, biochemical abnormalities, and phenotypic heterogeneity in peroxisome biogenesis disorders. J. Med. Genet. 36, 779-781.

Disease Models & Mechanisms DMM

671

RESEARCH ARTICLE

Drosophila peroxisome assembly

Shimozawa, N., Nagase, T., Takemoto, Y., Funato, M., Kondo, N. and Suzuki, Y. (2005). Molecular and neurologic findings of peroxisome biogenesis disorders. J. Child Neurol. 20, 326-329. Steinberg, S., Chen, L., Liumei, W., Moser, A., Moser, H., Cutting, G. and Braverman, N. (2004). The PEX Gene Screen: molecular diagnosis of peroxisome biogenesis disorders in the Zellweger syndrome spectrum. Mol. Genet. Metab. 83, 252-263. Steinberg, S. J., Dodt, G., Raymond, G. V., Braverman, N. E., Moser, A. B. and Moser, H. W. (2006). Peroxisome biogenesis disorders. Biochim. Biophys. Acta 1763, 1733-1748. Titorenko, V. I. and Rachubinski, R. A. (2001). The life cycle of the peroxisome. Nat. Rev. Mol. Cell Biol. 2, 357-368. Titorenko, V. I., Smith, J. J., Szilard, R. K. and Rachubinski, R. A. (1998). Pex20p of the yeast Yarrowia lipolytica is required for the oligomerization of thiolase in the cytosol and for its targeting to the peroxisome. J. Cell Biol. 142, 403-420. Tower, R. J., Fagarasanu, A., Aitchison, J. D. and Rachubinski, R. A. (2011). The Peroxin Pex34p functions with the Pex11 family of peroxisomal divisional proteins to regulate the peroxisome population in yeast. Mol. Biol. Cell 22, 1727-1738. van den Bosch, H., Schutgens, R. B. H., Wanders, R. J. A. and Tager, J. M. (1992). Biochemistry of peroxisomes. Annu. Rev. Biochem. 61, 157-197.

Vizeacoumar, F. J., Torres-Guzman, J. C., Bouard, D., Aitchison, J. D. and Rachubinski, R. A. (2004). Pex30p, Pex31p, and Pex32p form a family of peroxisomal integral membrane proteins regulating peroxisome size and number in Saccharomyces cerevisiae. Mol. Biol. Cell 15, 665-677. Wanders, R. J. A. and Waterham, H. R. (2006). Biochemistry of mammalian peroxisomes revisited. Annu. Rev. Biochem. 75, 295-332. Wilson, G. N., Holmes, R. D. and Hajra, A. K. (1988). Peroxisomal disorders: clinical commentary and future prospects. Am. J. Med. Genet. 30, 771-792. Yan, M., Rachubinski, D. A., Joshi, S., Rachubinski, R. A. and Subramani, S. (2008). Dysferlin domain-containing proteins, Pex30p and Pex31p, localized to two compartments, control the number and size of oleate-induced peroxisomes in Pichia pastoris. Mol. Biol. Cell 19, 885-898. Yik, W. Y., Steinberg, S. J., Moser, A. B., Moser, H. W. and Hacia, J. G. (2009). Identification of novel mutations and sequence variation in the Zellweger syndrome spectrum of peroxisome biogenesis disorders. Hum. Mutat. 30, E467-E480. Zhang, R., Chen, L., Jiralerspong, S., Snowden, A., Steinberg, S. and Braverman, N. (2010). Recovery of PEX1-Gly843Asp peroxisome dysfunction by small-molecule compounds. Proc. Natl. Acad. Sci. USA 107, 5569-5574.

Disease Models & Mechanisms DMM

672

dmm.biologists.org